AI Observability
Analyze AI cost and models
Compare model usage, token volume, latency, errors, and estimated cost.
Cost workflow
- Open AI Observability > Cost.
- Choose the time range.
- Review total estimated cost, daily trend, provider split, and model split.
- Open Bill Shock if spend changed suddenly.
- Open Recommendations for routing or prompt optimization ideas.
Models workflow
- Open AI Observability > Models.
- Compare request count, tokens, latency, errors, and cost by model.
- Look for expensive models used on low-value workloads.
- Look for models with high error rate or slow response time.
Unpriced usage
If a model is not in the pricing table, tokens can still be recorded while cost shows as zero or unavailable. Treat those cost numbers as incomplete until pricing is configured.
Need more help?
Keep moving with the right next guide
If your team still runs into setup issues, empty dashboards, billing delays, callback failures, or alerting problems, continue with troubleshooting before re-running the entire onboarding flow.